Mercari API: from Monolithic to Microservices
0 likes442 views
The document discusses Mercari transitioning its monolithic API architecture to microservices. It describes how Mercari introduced an API gateway to front multiple independent services for things like listings, shipping, and notifications. It also details some of the technical stack changes including moving to Google Cloud Platform and Kubernetes for containers. Finally, it outlines some of the challenges of the transition, such as how to handle database access across platforms and testing of the many independent services.
Mercari API: from Monolithic to Microservices
- 1. Mercari API: from Monolithic to Microservices
- 2. Tonghui Li Tech Lead, Microservices Development @ummterry
- 3. Background
- 4. Service A Service B Mercari API
- 5. How to make a new service?
- 6. API Gateway Service A Service B Mercari API
- 7. API Gateway Service A Service B Offer Mercari API
- 8. API Gateway Service A Service B Offer Multiple services on a single endpoint SSL Termination DDoS Protection Common AuthZ/AuthN Mercari API
- 9. API Gateway Service A Service B Offer Mercari API
- 10. Technical Stack
- 11. API GatewayAuthority Service A Service B Sakura Offer Mercari API
- 12. API Gateway Google Cloud Load balancing Authority Service A Service B Sakura Offer Mercari API GCP Kubernetes Engine
- 13. API Gateway Google Cloud Load balancing Authority Service A Service B Sakura Offer Mercari API GCP Kubernetes Engine Cloud Resources Managed Services Container
- 14. API Gateway Google Cloud Load balancing Authority Service A Service B Sakura Offer Mercari API GCP Kubernetes Engine Cloud Resources Managed Services Container Over HTTP
- 15. API Gateway Google Cloud Load balancing Authority Service A Service B Sakura Offer Mercari API GCP Kubernetes Engine Cloud Resources Managed Services Container Over HTTP SSL Termination DDoS Protection Cloud Amor?
- 16. API Gateway Google Cloud Load balancing Authority Service A Service B Sakura Offer Mercari API GCP Kubernetes Engine Cloud Resources Managed Services Container Over HTTP Routing to microservices Protocol tranformation (HTTP to gRPC) Common logging & Tracing Request buffering SSL Termination DDoS Protection Cloud Amor?
- 17. API Gateway Google Cloud Load balancing Authority Service A Service B Sakura Offer Mercari API GCP Kubernetes Engine Cloud Resources Managed Services Container Over HTTP Routing to microservices Protocol tranformation (HTTP to gRPC) Common logging & Tracing Request buffering SSL Termination DDoS Protection Cloud Amor? Common AuthZ/AuthN
- 18. API Gateway Google Cloud Load balancing Authority Service A Service B Sakura Offer Mercari API GCP Kubernetes Engine Cloud Resources Managed Services Container Over HTTP Routing to microservices Protocol tranformation (HTTP to gRPC) Common logging & Tracing Request buffering SSL Termination DDoS Protection Cloud Amor? Common AuthZ/AuthN Managed DB
- 19. How to Break Down existing API?
- 20. • Start from high-priority functionality
- 21. • Start from high-priority functionality • listing, search, purchase, shipping, notification, cs-tool, etc.
- 22. • Start from high-priority functionality • listing, search, purchase, shipping, notification, cs-tool, etc. • Split out necessary microservices
- 23. • Start from high-priority functionality • listing, search, purchase, shipping, notification, cs-tool, etc. • Split out necessary microservices • listing: master, user, item, etc. • shipping: master, user, item, address, etc. • …
- 24. • Start from high-priority functionality • listing, search, purchase, shipping, notification, cs-tool, etc. • Split out necessary microservices • listing: master, user, item, etc. • shipping: master, user, item, address, etc. • … • Come with up overall design
- 25. Listing
- 26. Shipping
- 27. Notification
- 28. Challenges …
- 29. Challenges: DB DB is in Sakura and we want to run microservices in GCP, but we don’t want to expose direct DB access over public internet • Deploy DB access services in Sakura • Migrate DB to GCP / new Tokyo DC
- 30. Challenges: Testing With so many services, how do we test them? • Automated regression testing for each service • Reduce required manual tests
- 31. Challenges: Transactions In monilith api, we often do transactions across different services: Solution: refactor and ensure Eventual Consistency
- 32. Thank you!